Influence of fracture criteria on the outcome of a randomized trial of therapy

Abstract

Based on a presumption that the morphometric approach to vertebral fracture assessment is beset by false positive deformities, it has been suggested that more stringent criteria should be used to asses vertebral fracture outcomes in clinical trials of osteoporosis therapies. We applied a variety of criteria in the reanalysis of a randomized trial of sodium fluoride therapy for women with established osteoporosis. Although progressively more severe criteria reduced the cumulative incidence of a new vertebral fracture in both those receiving a placebo and the fluoride-treated patients, differences between the two groups were not magnified and none of the suggested approaches produced a statistically significant result. These findings indicate that the false positive rate associated with morphometric assessment of vertebral fractures is not so great as supposed from theoretical considerations that assume independence among the measurements. Our findings also provide some reassurance that differences in the criteria used to define a vertebral fracture are not the controlling influence in the likelihood that a particular clinical trial will find a result favoring one treatment over another.